A Weld Line Detection Method Based on 3D Point Cloud for Automatic NDT

Abstract

Abstract The quality of welding is often checked by ultrasonic waves. Manual testing is costly and inefficient, and manual testing is not possible in some extreme environments. Automatic non-destructive testing (NDT) technology uses robots to carry ultrasonic devices for automatic detection. Machine vision is one of the important methods to achieve navigation, that is, capturing the weld line through the camera, planning the optimal path through visual analysis and processing, and also based on structured light. Although the navigation method can solve the problem of rust and stain to a large extent, it is less robust in dealing with problems such as rust, light interference and stain. This paper proposes a navigation method based on 3D point cloud, which can effectively improve its robustness.